What problem does it solve?
Running AI research workflows often requires manual terminal monitoring to track experiment progress, review scores, and pipeline milestones, leading to wasted time and missed critical updates.
Core Features & Use Cases
- Zero-overhead silent operation: If no Feishu configuration is present, the skill does nothing and leaves all existing workflows completely unaffected.
- Push notification mode: Sends fire-and-forget webhook alerts for key events like experiment completion, review scoring, and pipeline finish, with color-coded cards for different event types.
- Interactive bidirectional mode: Allows users to approve, reject, or provide custom input directly from Feishu messages for checkpoint steps, with automatic fallback if no reply is received within the timeout.
- Use Case: While running a long-running vertebrae segmentation experiment, you will receive a Feishu alert as soon as it completes with the full results table, no need to constantly check the terminal output.
Quick Start
Use the feishu-notify skill to send a custom status update about the current experiment progress to Feishu.
